We look at a comparison between a few ST analyses on the HER2-positive breast tumors provided in Andersson et al. (2021). The data consists of 7 patients each with 3-6 slices. So far, this analysis is only on patients ‘A’ and ‘B’. The following analyses were performed:

  1. BayesTME applied on each slice. We enforce 6 cell types and lambda of 10000. We sample 1000 times, with 2000 burn in steps and a thinning of 5.

  2. BayesTME applied on each slice. We enforce 6 cell types and lambda of 1000000. We sample 1000 times, with 2000 burn in steps and a thinning of 5.

  3. Latent Direchlet Allocation applied (LDA) on each slice, based on sklearn.decomposition.LatentDirichletAllocation. We enforce 6 cell types.

  4. Latent Direchlet Allocation applied on each patient. We enforce 6 cell types.

  5. The baseline analysis performed by Andersson et al. (2021). Nearest neighbors/hierarchical clustering is performed on gene expression at the patient+cohort level. This was used to set the number of cell types in each slice.

This set of figures shows the spot probability by cell type for each of the 5 methods.

This set of figures considers only the BayesTME and slice-based LDA. For each BayesTME cell type, we compute the 5 highest marker genes. We then analyze gene-topic probability from the slice-based LDA, to assess whether both methods have similar gene-latent space relationships.

## Warning: Removed 12 row(s) containing missing values (geom_path).

## Warning: Removed 12 row(s) containing missing values (geom_path).